The repertoire of ICE in prokaryotes underscores the unity, diversity, and ubiquity of conjugation

Abstract

Horizontal gene transfer shapes the genomes of prokaryotes by allowing rapid acquisition of novel adaptive functions. Conjugation allows the broadest range and the highest gene transfer input per transfer event. While conjugative plasmids have been studied for decades, the number and diversity of integrative conjugative elements (ICE) in prokaryotes remained unknown. We defined a large set of protein profiles of the conjugation machinery to scan over 1,000 genomes of prokaryotes. We found 682 putative conjugative systems among all major phylogenetic clades and showed that ICEs are the most abundant conjugative elements in prokaryotes. Nearly half of the genomes contain a type IV secretion system (T4SS), with larger genomes encoding more conjugative systems. Surprisingly, almost half of the chromosomal T4SS lack co-localized relaxases and, consequently, might be devoted to protein transport instead of conjugation. This class of elements is preponderant among small genomes, is less commonly associated with integrases, and is rarer in plasmids. ICEs and conjugative plasmids in proteobacteria have different preferences for each type of T4SS, but all types exist in both chromosomes and plasmids. Mobilizable elements outnumber self-conjugative elements in both ICEs and plasmids, which suggests an extensive use of T4SS in trans. Our evolutionary analysis indicates that switch of plasmids to and from ICEs were frequent and that extant elements began to differentiate only relatively recently. According to the present results, ICEs are the most abundant conjugative elements in practically all prokaryotic clades and might be far more frequently domesticated into non-conjugative protein transport systems than previously thought. While conjugative plasmids and ICEs have different means of genomic stabilization, their mechanisms of mobility by conjugation show strikingly conserved patterns, arguing for a unitary view of conjugation in shaping the genomes of prokaryotes by horizontal gene transfer.

Figure 1. Scheme of some essential interactions in the process of ICEs movement.

The integration/excision reaction (1) occurs by recombination across two recombination sites (yellow squares) located at the termini of the inserted element. As a result, a circular (most commonly non-replicating) DNA molecule is produced. Conjugation (2) is carried out by mobility systems. The relaxase (red circle) cleaves a specific site within oriT, and this step starts conjugation. The DNA strand that contains the relaxase protein covalently bound to its 5′-end is displaced by an ongoing conjugative DNA replication process (dotted lines). The relaxase interacts with the T4CP (green oval) and then with other components of the T4SS (blue rectangle). As a result, the relaxase-DNA complex is transported to the recipient cell . Since ICEs are supposed not to replicate autonomously, the process terminates necessarily by integration of the transferred DNA circle in the recipient chromosome.

Figure 2. Methods and results of the identification pipeline.

Upper. Diagram of the method used in the detection of the major representatives of the conjugation machinery: VirB4, T4CP, relaxases (rel) and the T4SS type-specific products. From expert datasets for the different proteins, we made multiple alignments and built HMM profiles that were used to scan chromosomes and plasmids. The numbers correspond to the number of hits. Lower. We then clustered co-localizing genes that are found within a maximum distance of 60 ORFs. A cluster containing a VirB4, a T4CP and a relaxase is considered as a putative conjugative system (CONJ). A cluster containing a VirB4 (plus or minus T4CP) but lacking a relaxase is considered as a putative protein-exporting T4SS. A cluster containing a relaxase but lacking a VirB4 is considered as a mobilizable element (MOB). The table shows the number of each type of clusters.

Figure 3. Percentage of genomes containing T4SS in function of genome size.

Red: MOBless T4SS clusters T4SS+MOB-, Blue: ICE, Green: conjugative plasmids.

Figure 4. Distribution of ICEs and conjugative plasmids in the different groups of the VirB4 phylogenetic tree.

(See also Figure 6.) The VirB4 families correspond to the four previously described proteobacterial MPF types (T, G, F and I) plus four additional families, associated with either host phyla (bacteroidetes, cyanobacteria) or mixtures of phyla (FA = firmicutes plus actinobacteria) or FATA (firmicutes, actinobacteria, tenericutes and archaea).

Figure 5. Distribution of the different MOB families among clades.

(See also Figure 6.) The figure shows the percentage of each MOB type (color code at the right) associated to each MPF type.

Figure 6. Phylogenetic analysis of a non-redundant subset of VirB4 proteins (only one protein per cluster of proteins >90% identical was used).

Top: phylogeny of VirB4 with groups cartooned together. Bootstrap values above 75% are depicted. Bottom: phylogeny of VirB4 within MPF_T (in gray an outgroup to root the tree) with an indication if the gene is encoded in a chromosome (black circles) or in a plasmid (red).

Figure 7. Analysis of phylogenetic associations between conjugative systems.

The x-axis represents the pairwise evolutionary distances between VirB4 proteins (as taken from the tree in Figure 6). The histogram represents the counts in each bin of evolutionary distance (e.g. 5488 comparisons for proteins at an evolutionary distance less than 1). The line represents the frequency with which each of the two VirB4s at a given evolutionary distance belong to the same type of replicon, i.e. a plasmid or a chromosome. The line is a spline fit on the graph of pairwise evolutionary distances versus the frequency with each the two VirB4 are in the same type of replicon (i.e. both chromosomal or both plasmidic). For closely related sequences (smallest distances), the y-value is close to 100%, indicating that sequences belong to the same type of replicon. When the distance increases, the variable falls to 50%, indicating that the probability of finding a pair of VirB4 at these evolutionary distances in two chromosomes or in two plasmids is the same to that of finding them in a pair constituted by one ICE and one conjugative plasmid.